Introduction

Soil is the most important exhaustible natural resource because it is not possible to return if it is destroyed or lost through anthropogenic activities. Soils are mostly eroded through the different processes like sheet erosion, tunnel erosion, rill and formation of gully which are the extreme forms of degradation of land resources (Ghosh et al. 2016). It is found that soil erosion costs the United State economy between US $ 30 billion to US $ 44 billion annually. In Indonesia this cost is nearly US $ 400 million per year in Java alone (Morgan 2005).The weathered materials of rocks are transported by a particular process are popularly familiar as erosion. There are numerous agents which are associated with erosion these are water, glacial, wind, waves etc. There are two clear cut phases of soil erosion these are the process of detachment of particles from top soil and transportation of the same materials by active agent like water and wind (Bhandari et al. 2015). Soil degradation is one of the most important elements of land degradation by which the physical, biological as well as chemical environment are degraded. Chemical degradation is mainly associated with the process of leaching, exploitative cropping system and poor irrigation practices (Chemeda 2007). Leaching process of chemical degradation is common in most of the tropical and sub tropical countries. Biological degradation of soil means the decreasing tendency of biological activity. When the amount of vegetation of a specific area is destroyed then the biological and ecological association of a soil is declined. Physical degradation is the amalgamation of numerous interrelated processes and it is included with chemical and biological degradation process (Chemeda 2007). Land degradation from soil loss is a monotonous issue in most of the developing countries like India and it is estimated that about 3.975 million hectors land in India have been degraded due to soil loss (Ghosh et al. 2010). Land degradation from soil loss is a common issue and it is growing rapidly. There are numerous works on land degradation due to soil loss (Foster and Meyer 1972; Renard et al. 1997; Millward and Mersey 1999; Van der Knijff et al. 1999; Sadiki et al. 2004; Hasim et al. 2005; Ghosh and Guchhait 2012; Prasannakumar et al. 2012; Aiello et al. 2015; Belasri and Lakhouili 2016).

Several researchers identified that GIS and Remote Sensing is the most reliable and dependable tools in the measurement of soil erosion through different empirical and semi empirical models (Irvem et al. 2007; Terranova et al. 2009; Kouli et al. 2009; Demirci and Karaburun 2012; Ganasri and Ramesh 2016). This is the arrangement of empirical/semi-empirical nature and process based model (Biswas and Pani 2015).

In RUSLE, the rainfall runoff factor of the original USLE (Universal Soil Loss Equation) was replaced by the rainfall erosivity factor (Renard et al. 1997; Dutta 2016; Bera 2017).The Universal Soil Loss Equation (RUSLE) was used for predicting the long term average annual soil erosion on the basis of different factors which is directly and indirectly related with soil loss in a specific area. The RUSLE is an empirical equation that computes the average annual soil loss in tons /ha/year. The RUSLE is the earliest quantitative soil loss models, is relatively easy and acceptable and that has been applied in various region of the world in modified form (Roche 1954; Yin et al. 2006; Angima et al. 2003). But its initial function and formula changes and its modified form applied in the Asia, Africa and Australia’s environment (Angima et al. 2003).

The Revised Universal Soil Loss Equation (RUSLE) is one of the most popular and reliable method and it is based on ground based observation and integrated with Remote Sensing and GIS (Pandey et al. 2007; Sharma 2010; Pal and Samanta 2011; Shit et al. 2015; Biswas et al. 2015; Samanta et al. 2016a, b; Pal and Shit 2017). This model have been use widely for its simple function and available data function (Jain and Kothyari 2000; Bhattarai and Duttta 2007; Pandey et al. 2007; Sinha and Joshi 2012; Jiang et al. 2015; Balasubramani et al. 2015, Biswas et. al. 2015).

In RUSLE model, the amount and direction of slope and characteristics of aspect are applied for mainly the purpose of segmentation (Lewis et al. 2005; Tetford et al. 2017; Tilahun et al. 2018; Wijesundara et al. 2018). Integration of statistics and GIS is one of the important synthetic tools for predicting the present status of soil erosion as well as the prediction about the future scenario of soil loss. Selected factors have to be assign according to their weight then soil erosion hazard have been identified with the help of Z score (Rahman et al. 2009). This model can easily indent the nature of eroded materials which is deposited through sediment transportation (McCool et al. 2004).

In RUSLE with integration of GIS the sediment yields have been measured through the Sediment Assessment Tool for Effective Erosion Control (SATEEC) (Lim et al. 2005). The amount of potential soil loss through RUSLE model in GIS framework can identify the vulnerable areas which is very much susceptible for development of rill, gully etc. (Fagbohun et al. 2016). The vulnerable land use classes can be estimated from the RUSLE model in GIS environment for mainly in planning purposes (Balasubramani et al. 2015). Beside this several researchers applied RUSLE model for accounting the soil loss in numerous fields: (Mccool et al. 1987; Moore et al. 1992; Busacca et al. 1993; Renard et al. 1994; Yoder and Lown 1995; Liu et al. 2000; Wang et al. 2000; Stolpe 2005; Prasannakumar et al. 2012; Mallick et al. 2014; Predeep et al. 2015; Mondal et al. 2016).

MCDA (Multi Criteria Decision Analysis) is one of the most reliable toll for identifying the vulnerable areas of soil erosion and their associated surface lowering with incorporating the all phenomena that is related with soil loss in precise manner (Pal 2016). Samanta et al. (2016a, b) has been used the RUSLE in GIS environment for indentifying the vulnerable areas with the help of soil erosion susceptibility mapping as well as and measures the soil conservation planning in the soil erosion vulnerable areas. Hembram and Saha (2018) established that the morphometric attributes of a drainage basin can play a significant role in relation to soil erosion and prioritization of sub-watersheds with the help of fuzzy AHP and compound factor can reduce the reduce the rate of soil erosion within this basin. Assessment of soil erosion susceptibility is essential through soil erosion susceptibility mapping is essential in the soil erosion vulnerable areas and take initiative through proper land use and management practices (Pal and Debanshi 2017). Gayen and Saha (2017) used weights from evidence and evidential belief function for preparing the soil erosion susceptibility mapping which established in ground reality with adequate accuracy.

When the adequate primary information regarding the rainfall in storm period is unavailable then the TRMM data is one of the most important reliable sources for predicting the rainfall and runoff erosivity factor (MJ mm/ha/hr/year) because it provides the high resolution precipitation datasets which almost similar to ground reality (Heiblum et al. 2011; Dutta et al. 2015).

The availability the quantitative data with good quality is not sufficient in most of the developing countries hence the application of such type model is limited. Every empirical and semi empirical model have some unique function so one single model is not enough to fulfill the objectives. Determinations of RUSLE variables are necessary but the acquiring and fitting of observed data is time dependent and become costly. The major objective of this work is to estimate the average annual soil loss and to highlight the future potential soil loss risk zones with spatial coverage of Arkosa watershed for future which may help to take suitable remedies for sustainable land use practices.

Location of the study area

Arkosa is a right bank and important tributary of river Dwarkeswar. It originates near Hura of Puruliya District and meets with River Dwarkeswar near Ramnagar village near the town of Bankura Sadar of Bankura District. It is bounded by latitude 23° 9′49.49"N to 23°20′24.18"N and longitude 86°37′48.30"E to 86°54′53.24"E in corresponding area of 348.5 Km2 (Fig. 1). According to the watershed atlas of India this study area belongs to the watershed codification of 2A2C8. The river originates from near Tilaboni hill of Puruliya district and enters the Bankura district in Chhatna C.D. block (Chakrabortty et al. 2018). It passes through Bankura town and enters into the southeastern part of Purba Bardhaman district. The River Dwarkeswar ultimately leaves behind into Hooghly district.

Fig. 1
figure 1

Location of the study area

Full size image

Data used

Keeping in the view of objectives of this work the following datasets have been use to continue this work, these are: Topographical map by Survey of India, Landsat 8 OLI sensor satellite data by United State Geological Survey (http://www.usgs.gov/), Shuttle Radar Topographic Mission (SRTM) Digital Elevation Model (DEM) by United State Geological Survey (http://www.usgs.gov/), Soil Texture map by NBSS&LUP Kolkata, Rainfall records from rain gauge station etc. the R (rainfall and runoff erosivity) factor raster has been created from the observed rainfall records in storm rainfall periods. The K (soil erodibility) factor raster has been estimated from the soil texture records and its chemical properties that is established by NBSS&LUP Kolkata. The LS (slope length and steepness) factor has been estimated from the SRTM MEM with considering the nature of flow accumulation and characteristics and nature of slope in GIS environment. The vegetation algorithm raster like NDVI (Normalized Difference vegetation Index) form Landsat 8 OLI sensor satellite data has been taken into consideration for estimating the C (cover and management) factor in GIS environment. The P (support practice factor related to slope direction) factor raster has been estimated from the percentage slope in SRTM DEM and observed support and management practices within this region during the field visit.

Materials and method

Catchment wise soil erosion is estimated through numerous methods, these models have been classified according to their nature i.e., empirical model, semi-empirical model, physical based model and conceptual model. Universal Soil Loss Equation (USLE) (Wischmeier and Smith 1978) and Revised Universal Soil Loss Equation (RUSLE) (Renard et al. 1997) are included in the empirical model. Modified Universal Soil Loss Equation (MUSLE) (Williams and Bernd 1977), Morgan, Morgan, and Finney (MMF) Model (Morgan et al. 1984) and Revised Morgan Morgan Finney (RMMF) Models are included with semi-empirical model. Water Erosion Prediction Project (WEPP), Soil and Water Assessment Tool (SWAT), Agricultural Non Profit Source Pollution Model (AGNPS), European Soil Erosion Model (Morgan et al. 1998, 1999; Kinnell 1999) are included with the physical based model. Sediment Concentration Graph (Johnson 1943), Renard-Laursenn Model (Renard and Laursen 1975), Unit Sediment Graph (Rendon-Herrero 1978), Instantaneous Unit Sediment Graph (Williams 1978), EMSS (Vertessey et al. 2001), HSPF (Johanson et al. 1980), IQQM (DLWC 1999), LASCAM (Viney and Sivapalan 1999) and SWRRB (USEPA 1994) are included with conceptual model.

In this study Revised Universal Soil Loss Equation (RUSLE) has been taken into consideration to estimate the average annual soil loss of Arkosa watershed. Application of RUSLE model in GIS framework was vastly used even in rugged topography, tropical forests and the watershed with a steep slope (Samanta et al. 2016a, b) and even in rugged topography. Universal Soil Loss Equation (USLE) is generally accepted for its simplicity but it is difficult to measure rainfall-runoff factor in a single framework. So in RUSLE in this factor replaced as a rainfall and runoff erosivity factor (Fagbohun et al. 2016). So different researcher of the world has been used RUSLE in GIS environment to fulfill the research objectives in less quantity data sets but an adequate accuracy (Zhihua et al. 2002; Lu et al. 2004; Fu et al. 2006; Yue-Qing et al. 2008; Chen et al. 2011; Kumar and Kushwaha 2013).

The factors that are used in RUSLE acquired from rain gauge station, soil data, DEM and satellite image etc. The five thematic layers have been taken into considerations as input of RUSLE model in GIS environment for estimating average annual soil loss of the Arkosa watershed. Raster calculator of Spatial Analysis Tool has been used for creating each layer in ArcGIS 10.4 environment. In RUSLE the five thematic layers are multiplied in the following equation (Fig. 2):

Fig. 2
figure 2

Methodology flow chart

Full size image
$$A=R \times K \times LS \times C \times P$$

Where

A = Average annual soil erosion (ton/ha/year)

R = rainfall and runoff erosivity factor (MJ mm/ha/hr/year)

K = soil erodibility factor (ton/ha)

LS = slope length and steepness factor

C = cover and management factor

P = support practice factor related to slope direction

Principal Component Analysis (PCA) has been used for identifying the importance of each factor in an index. Therefore, to fulfill this criterion of overall area has been divided in to 2 Km and 2 Km grid and their central values (89 points of each thematic layer and total 445 points) have been taken into consideration. Then the combined factor indexes indices of different factors which have been calculated for showing the combine impact of all factors in soil loss. This index has been calculated from the following formula:

$$\frac{{\frac{1}{n}\mathop \sum \nolimits_{{i=1}}^{n} (xi - \bar {x})~(yi - \bar {y})}}{{\frac{1}{n}\mathop \sum \nolimits_{{i=1}}^{n} {{(x - \bar {x})}^2}}}$$

For predicting the future average annual soil loss, rainfall and runoff erosivity factor, soil erodibility factor and slope length and steepness factor have been taken into consideration (Biswas and Pani 2015).

Results and discussion

R factor

The rainfall and runoff erosivity factor specify the erosive capacity or erosive power which take place in the storm rainfall period (Pal and Shit 2017). It indicates the average annual storm rainfall value and it is associated with possible soil erosion of a particular region (Das et al. 2018). The R factor calculated from the annual average observed rainfall data (Table 1).There is a positive relationship between the amount of rainfall and runoff, though there is a direct and indirect impact of slope and it is acted as a determining element of runoff. The R factor of a specific region is expressed as MJ mm ha-1 year-1 (Wischmeier and Smith 1978).

$$R~=~38:5~+~0:35 \times \Pr$$

where, R is rainfall and runoff erosivity factor and Pr is the average weekly precipitation (mm) in storm rainfall period. In sub-tropical monsoon dominated countries like India, the monsoon period is considered as storm rainfall event for estimating the rainfall and runoff erosivity factor.

Table 1 Average rainfall data from rain gauge station
Full size table
Fig. 3
figure 3

R factor of Arkosa Watershed

Full size image

The R value of this region ranges between 58.347 to 58.778 MJ mm ha-1 year-1. In this region, the highest value of R factor concentrated in eastern and southern portion (Fig. 3). The lowest R value is intense in the northwestern part of this watershed. In other portion, the value of R factor is moderate in nature.

K factor

Soil erodibility indicates the capacity to loss the soil and it depends upon various chemical and physical characteristics properties of the soil (Pe´rez-Rodrı´guez et al. 2007). Though the mineral components and morphological characteristics of various parameters of soil, take part in a significant function in erosion susceptibility (Biswas and Pani 2015). The K factor signifies the combining effort of rainfall, runoff and amount of infiltration that is associated to soil loss in a storm event (Renard et al. 1997). K factor emphasis on the erosion vulnerability and amount of runoff (Pal and Shit 2017). A general soil texture map prepared by NBSS&LUP, Kolkata has been used for estimating the soil erodibility factor. The K factor was identified with the help of Soil erodibility monograph (Wischmeier and Smith 1978) by considering soil composition and organic content. Soil types are classified into four textural classes: Fine (120.66 Km2), Fine Loamy (121.22 Km2), Fine Loamy-Coarse Loamy (1.67 Km2) and Gravelly Loam (104.90 Km2).

The soil K factor was calculated using this formula (Wischmeier and Smith 1978):

$$K~=~2.1 \times ~10~-~6M1.4~ \times ~(12~-~OM)~+~0.0325~ \times ~(P~-~2)~+~0.025~ \times ~(S~-~3)$$

Where, K = soil erodibility factor (ton ha-1 unit of R); M = (% silt +% fine grained sand) (100 − % clay); OM = organic matter in percentage; P = permeability; and S = structural classess (Pal and Shit 2017).

The K factor of this region ranged between 0 and 0.33 ton/ha. The lowest K factor value (0 to 0.14) mainly concentrated in the western part of the watershed (Fig. 4). The moderate K factor value (0.14 to 0.19) also concentrated in the northern portion and the higher value (0.19 to 0.23) concentrated only in the eastern portion. The other portion of this watershed are belongs to very high K factor value (0.23 to 0.33).

Fig. 4
figure 4

K factor of Arkosa Watershed

Full size image

LS factor

In RUSLE, LS factor have been generated with the integration of slope length factor (L) and the steepness factor (S). It is also recognized as topographic factor (Pal and Shit 2017). There are two processes to estimate the LS factor; these are direct field measurement and Digital Elevation Model (DEM). The slope and flow accumulation grid (Figs. 5, 6) was extracted from the Digital Elevation Model (DEM), and then these two parameters employed for estimate the LS factor in GIS environment. The integration of slope length factor (L) and the steepness factor (S) create the LS factor in GIS environment with considering the following equation (Moore and Burch 1986):

Fig. 5
figure 5

Slope of Arkosa Watershed

Full size image
Fig. 6
figure 6

Flow accumulation of Arkosa Watershed

Full size image
$$\begin{aligned}LS &=(Flow~Accumulation \times cell~size/22.13)0.4 \\ & \quad \times (Sin~[Slope~grid~ \times ~0.01745]~/~0.0896,~1.4)~ \times ~1.4 \end{aligned}$$

The model builder tool in GIS environment use to considerate for estimating the LS raster of this study area. The equation of LS grid was (Pal and Shit 2017):

$$\begin{aligned}LS &=Pow~([Flow~Accumulation~Grid] \times ~10/~22.13,0.4) \\ & \quad\times Pow~(Sin~[Slope~grid~ \times ~0.01745)/0.0896,~1.4) \times 1.4\end{aligned}$$

Where Pow denotes for Power equation in GIS environment, Flow Accumulation created in GIS environment from DEM for estimating the flow in grid format, and cell size is the length of a raster cell. The L and S factors were worked out from a DEM of the Arkosa Watershed. The LS factor values ranges between 0 and 3.97. These have been classified into different LS factor threshold like as 0, 0.05, 0.11, 1.75 and 3.97.

The low LS factor (0–0.05) values found in the far away from the major and minor streams of this watershed. The moderate LS factor (0.05–0.11) values mainly concentrated only few portion of this watershed. The high LS factor (0.11–1.75) values mainly concentrated in the nearest to the major and minor streams and the very high LS factor (1.75–3.97) values mainly concentrated basically where the major and minor streams are located of this watershed (Fig. 7).

Fig. 7
figure 7

LS factor of Arkosa Watershed

Full size image

C factor

In USLE the C factor was mainly associated with the field observation but in RUSLE there are four sub factors which are associated and considered for accounting this amount. These are specified land use of this area, the status of vegetation cover, the characteristics and cover of the land surface and the roughness of the surface. C factor is one of the important dimensionless factors that indicate the amount of soil loss directly related to the vegetation cover. This factor is the proportion of soil loss by the areas of defensive vegetation cover (Donahue et al. 1972). Here Landsat 8 OLI satellite data have been used to generate NDVI map of the study area.

Normalized difference vegetation index (NDVI) is one of the most dependable and famous method to estimate the vegetation cover (Pal et al. 2018). NDVI generally estimated from the following equation by Rouse et al. (1974):

$$NDVI~=~(NIR~ - ~{\text{Red}})/(NIR~+\operatorname{Red})$$

It is mainly ranged with Negative one (− 1) to positive one (+ 1). Negative one (− 1) to zero (0) represents the water body to saturated or moist soil and zero (0) to positive one (+ 1) represents bare soil surface to completely developed vegetation cover (Pal et al. 2018). NDVI values of the study area ranged from − 0.17 to 0.44 (Fig. 8). After the generation of NDVI thematic layer the following equation have been taken into consideration for estimate the C factor:

Fig. 8
figure 8

Normalized difference vegetation index of Arkosa Watershed

Full size image
$$C~factor~=~1.02~-~1.21~ \times ~NDVI$$

C factor was created in using the specified equation that included field information of land cover (Wischmeier and Smith 1978; Renard 1997; Renard et al. 1997). The values of C factor in this study area varied between 0.48 and 1.23. These have been classified into different C factor threshold like as 0.48, 0.84, 0.90, 0.94 and 1.23. There is a positive relationship between the values of NDVI and C factor. The lowest values of C factor (0.48–0.84) mainly found low in the western, northern and southeastern part of this watershed in some small pockets (Fig. 9). The moderate C (0.84–0.90) values mainly concentrated in the northern, eastern, southern and southwestern portion of this watershed. The high C (0.90–0.94) values concentrated in the northeastern, central portion and the very high (0.94–1.23) values mainly concentrated in the northern, western and central portion of this watershed.

Fig. 9
figure 9

C factor of Arkosa Watershed

Full size image

P factor

The P value indicates the percentage of soil loss form a field with support practices (Pandey et al. 2007; Blanco-Canqui and Lal 2008; Blanco and Lal 2010; Brady and Weil 2012). The P values generally vary from 0 to 1. The highest P value indicates the bare surface lacking any support practices. Maintaining the correspondence between living and dead vegetation and emphasis upon the conservation tillage could reduce the rate of erosion (Bancho and Lal 2008). This factor generally related with the different type of erosion in different cultivated land where different types of cropping practiced have been taken into consideration (Pal and Shit 2017). For minimizing the soil loss in a watershed, use of multi practice is more useful than a mono practice. In this case support practice like contouring, strip cropping and terracing must be encompasses (Bancho and Lal 2008). Apart from this field bunding kind of management practices may be useful for the region where intensive subsistence agricultural practices are going on likewise Arkosa watershed (Fig. 10). Alteration of flow pattern could influence the soil erosion by support practice (Renard and Foster 1983). Basically this area belongs to the mono crop like paddy cultivation and others area is included with non paddy area. In this study area the P value ranged between 0.20 and 0.44. The lower value concentrated beside the streams. The higher value of P associated with the undulating topography as well as the non-agricultural areas (Fig. 11).

Fig. 10
figure 10

P factor of Arkosa Watershed

Full size image
Fig. 11
figure 11

Support practices adopted by the Villagers: a Photograph of Field Bunding for minimizing the removal of the top soil in the single crop Paddy cultivated land in Kashipur and its surroundings; b Photograph of Paddy dominated agricultural land where Field Bunding method has been adopted for controlling the soil erosion near the village of Natungram

Full size image

Estimation of soil erosion

Soil loss estimation through RUSLE in GIS framework not only for time consuming but also it can provide the result in a greater accuracy level. In manual method of soil loss estimation model is unable to predict the catchment wise soil loss status but in GIS framework is suitable for identifying the soil loss status. These study mainly emphasized upon the quantifying various erosion potential zones and to predict the future soil loss status. The correlation matrix between various factors shows a clear cut framework for understanding the importance of all factors in a single dimension (Fig. 12). Future management through suitable techniques by planers should take emphasis of vulnerable areas of present conditions as well as predicted erosion potential areas. The average annual soil loss in Arkosa watershed was estimated through RUSLE model (Fig. 13). The average annual soil loss in Arkosa watershed ranged from 0 to 10 tons/ha/year. Then it classified into different erosion threshold on the basis of a geometric interval for identifying different erosion classes i.e., 0, 2, 4, 8 and 10 tons/ha/year. The very high (8–10 tons/ha/year) soil loss mainly concentrated in the southeastern part of this watershed. The high (4–8 tons/ha/year) soil loss erosion zones were found in the southern and southeastern part of this watershed. The moderate (2–4 tons/ha/year) soil loss class concentrated in the northern, central and southeastern portion of this watershed. Most of the areas of this watershed belong to the low (0–2 tons/ha/year) soil loss zone. The spatial variations of different factors controlled the overall environment that the quantity of soil loss and severity are different in diverse environmental condition (Fig. 14). It mainly depends on the amount of storm rainfall, soil texture, vegetation cover and amount of slope (Table 2). The combined index of different factor shows the differences of influences on soil loss (Table 3). It is shown that the minimum (0–0.250) influences mainly found in the northwestern part of this watershed (Fig. 15). The minimum (0.749–0.999) influences found in the eastern and southwestern part of this watershed. Besides this, the major part of this watershed is associated with moderate (0.250–0.500) to high (0.500–0.749) influences of combine factor. The 1st component and 2nd component explain almost the 50 Percent (0.4925) of the variables (Table 4), by which we can say that there are a lot of differences found between various variables.The 1st and 2nd component is most important in this distribution because in those components the eigen value is greater than one (1) (Fig. 16). These differences between variables play a very important role in the overall distribution, while R and P factor is more dominant than other factors. Pair wise Compression Matrix shows the correlation between different factors. It is shown that the maximum correlation between R and C factor (Table 5). Here the partial and semi-partial correlation with Combined Factor Index shows the higher correlation among R factor and Combined Factor Index with important significance (Table 6). Besides this, the correlation of the other factor with Combined Factor Index is minimum because of the larger influences of R factor. We can at a glance says that the importance of different factors on soil loss through the help of the descriptive statistics (Table 7). Apart from this there are total 29 points have been selected randomly for collecting the primary information regarding the average annual soil loss (Fig. 17; Table 8). There is a highly positive correlation between the actual soil loss and estimated soil loss from regression analysis have been found which secure the validity of this present soil erosion model with actual ground reality (Fig. 18). The potential soil loss estimation map shows the future potentiality of soil loss which ranged between 0 and 16 tons/ha/year (Fig. 19; Table 9). This is classified into dissimilar potential soil erosion threshold for identifying different erosion classes i.e., 0, 4, 8, 12 and 16 tons/ha/year. The very high (12–16 tons/ha/year) potential soil erosion are mainly found in the eastern, northeastern and southeastern part of the Arkosa Watersahed. The high (8–12 tons/ha/year) amount of potential soil loss is mainly found in the eastern and northern part of this area. The spatial distribution of moderate (4–8 tons/ha/year) soil loss mainly concentrated in the southwestern and northwestern part of this watershed. The spatial distribution of marginal (0–4 tons/ha/year) soil loss mainly dominated in the western, southwestern and northwestern part of Arkosa Watershed (Table 9). In the overall study we can say that the quantity and spatial distribution of soil loss mainly controlled by the fluvial activity and the percentage of slope in the overall watershed. The percentage of the slope acts as a threshold in relation to soil loss but there is a direct as well as indirect relationship of the rainfall and runoff erosivity factor, soil erodibility factor, slope length and steepness factor, cover and management factor andsupport practice factor which are also related with the amount and direction of slope etc.

Fig. 12
figure 12

Correlating matrix between different factors of soil loss

Full size image
Fig. 13
figure 13

Average annual soil loss of Arkosa Watershed

Full size image
Fig. 14
figure 14

Field photograph on soil erosion: a Rill formation already started near Dalanbani village; b Sheet erosion and Rill development is going on near the Dalanbani village; c erosion of the top soil and exposed lateritic soil features are very common near Anandapur village and its surrounding region; d Root of the vegetation gets exposed due to severe soil erosion near Anandapur village

Full size image
Table 2 Average annual soil loss and its spatial coverage
Full size table
Table 3 Combined index value of different Factors
Full size table
Fig. 15
figure 15

Combined factor index of different factors

Full size image
Table 4 Eigen value of different Component
Full size table
Fig. 16
figure 16

Scree plot Eigen values

Full size image
Table 5 Pair wise compression matrix
Full size table
Table 6 Correlation of different factors against combined index
Full size table
Table 7 Descriptive statistics of different factors
Full size table
Fig. 17
figure 17

Location of the validation points

Full size image
Table 8 Validation of the study
Full size table
Table 9 Potential soil loss risk zones and its spatial coverage
Full size table
Fig. 18
figure 18

Regression between estimated and observed soil loss

Full size image
Fig. 19
figure 19

Potential soil loss risk zones of Arkosa Watershed

Full size image

Conclusion

Ground truth base primary investigation for creating database is time-dependent, costly and difficult but when it is applied with GIS framework it is capable to estimate the quantity of soil loss and its spatial distribution. This study mainly deals with the identification and demarcation of probable or potential soil erosion risk zones with the aid of RUSLE model and combined factor index method in GIS environment. This study provides a reliable prediction regarding soil erosion. From the above study, it has been observed that 17.73% of the total area included with high to very high annual average soil loss. But in the future potential soil loss, it goes to 35.978%. So the initial and leading work of the planer is to give emphasis within this area for reducing the quantity of eroded soil with the assist of suitable management practices. The low average annual soil loss of this area is 63.95% and in potential soil loss, it has been found about 37.849%. Here it has been observed that the rapid decline of low soil loss is very common within the region. That is the major problem of this watershed. The soil loss was mainly found in the highest elevation and high slope area where the runoff is maximum. On the other side, the moderate soil loss areas concentrated in the fallow and scrubland and the low soil loss area confirmed in the vegetative land. It is not possible to stop the soil erosion completely but proper or suitable land use management and suitable support practices can minimize the erosion rate and vulnerability of the top soil contained by the region. These counteractive measures can also retain the fertility of the top soil within the in-situ region which ultimately accelerates the productivity of the land in future.